haptor 3
Typology and reconstruction
Bernard Commie
1 Goneral considerations
‘The reconstruction of protolanguages is one of the most
traditional ass of kntoneaLcomparative ings, ined the
reconstruction of Proto lndo-Europeun i rightly considered one
Ot the trumps of netsnthsentry linguist science Although
typology can tac is ancy buck even frtber than stra
Hineuistis, only relatively reoenly that ft has butgeoned to
gin acceptance as @ flly-Redged sub-dscipine of tngusie,
‘What then, tthe postibe relaponcip between ese two su
lscipins? In order to answer this gueson, hs nesesary fo
hve somc understanding of what Kinds of questions Spology
{tes to answer, Whi difercot linguists might pve somewhat
Sitferen answers, Twi sseme tht the fling conse 2
Platatle subset of posible answers!
First, tppology” i eoncemed with the question of hat
consites 2 possibe human language. There are two distin but
interelated ways in which one might approsch this question, The
fist is empincal, te. we tey and place contains on posite
human languages by exauiting tnd senealizing sere what we
find in atested languages. The second is theoretical, on he
bss olinited aheenatons of ated languages we cmt
hypotheses sbout probable constrsnis on human languages,
wich we are then fee to fs against data from che nung.
Although the empincal a theoretical approaches are ofen
fonts, ren some antago, eet hey aly
‘eet diferent degrecs of emphas, since any generalization We
‘ate irom the set of observed languages to human langage In
feneral necessarily theoretical, dy theory desipned to be
IReasured agaist the facts of languages necessary empl,
_gENERAL. CONSIDERATIONS 1
language erreesent every language, wile the say oftypological generation to which no exception i nowa. Its
lvaye coneivable that sich a genealation might De over:
thrown by data om some se yet unsewrbed langage indeed
this i precely what happened withthe generatzaton about
Subjectypicaly precoding objects ia unmarked ord. ores
Srtch was until te reenly widely eld to be an absolute
Univeral Ta this spect, taportant to Bea ia mind that the
umber of linguist properis with which we are dealing is
fnmens. Given that we ave dealing with such large number of
Properics, Hs not Incomcctetie iat a fow auth propertis
Bight have very skewed atibutons, cvem give "te lange
umber of languages fo which we have aonss. If 2 puatve
‘solute universal has principled ass, Le can be shown *0
flow om. some more. general property of language oF
option, then irhas 3 beter hance of Being indeed exception.
ies wat rerpect to posible aman langvages, hough Renn this
doe ont geatonoe eholue rlabiiy,
‘How doer all ofthis reat to recomsricon in historia
compertve inguin? The end prodec of w reconstruction 8
‘Shatscteriation of a langusge that a hypothesized to have been
Spalcen inthe pas.* Ths esonsrocted anguage can be tated
{alist the sate typoloial generalizations as re applicable to
attested languages Thus, ws can sok ofa parealarrevnstucted
language whether is conten wit what we behve fo be
Constraints on possible human langusges, and ifthe answer is
negtve then we should serous teconsde the reconsrction
‘We can sho ask of @ particular reconstruction whether ts
Pnwible, though bore we are on more dangerows rowed since
Hmplusbe dacs not can the same ss impossible’, and there
are even some instances where typologaly implausible
Feconstuction might, for other reason, be fealy acetal
For itance, all Iaaguages of the ‘Kbossan fami of south
tester Ana have leks os pat of their regular phoneme
Taventoy, alfough cihs are Giherwise virally unknown 23
Part of tie reglsr phoneme ventory of languages. Standard
Procedures of reconstruction, in. partula the comparative
bnethod, would lend us to postulate Ces in Proo-Khosan. and
presumably mo one wold rjc! this rsoostrcton om grounds of
SSpolopicl inpauslity, piven that every descendant of Proto
oan har sks.
‘One limitation onthe application ofrpology, in pntiala of
implications) avers, tthe csiaton of constuctons
that any reconstruction is necesarypartal the evidence fs
stented lngungss enables bs to bal spa paral pstre of het
ancestor, bat there wil always be properties of the ancestor that
ire are unable to reconsract berause evidence for them has
teen fost from all descendant languages. For instance, examin
non of the moder Romance languages, Would not lead ws to
inclu that atin has asicase stem. Ths, & reconstruction
felk of things that the protodanguage had, but doesnot
ecesarly tell ws what things toot have. Tethore isan
Implictonsl universal py then and we rectmsruct a
Tamguape that has put lacks 9, tis could in principle be Sue 10
the attested lanplages’ having los all traces of But stl
febining enougit efdence Yo reconscic p when he proto
Tangxsge i fact had both p and
There jean important assumption implicit in the foregoing
namely tat thee hes een no change the range of potential
‘ppolopicl variation in human language since the time at which
‘Sith reconstructed languages are assed to have Been spoken
fn the preseat day. Infact forthe tme-depth at wbich we work
in mont rocostractions this is 2 plasible assumption. Proto
Indo-European, for ssance, ‘b Usally asumed to ave been
‘poten some 4500 to 60 year before the preseol, and mont
ther rconsracied proto languages involve similar or shallower
time-depis. This short penod compared. with the overall
Hitony ‘of human langage: the most conservative estimates
‘would date the Tetest possible time for the ong of Duman
Tnnguage to around 35.0 years belor the preseat (Le. the tme
‘ofthe start of the major migrations that ease Homo spies
‘host parts ofthe word), with 100.000 Years before the present
ing loser fo the consests date (2 de appearance of Homo
‘SSptce) Thus. if there are typological generaliations that
‘Suracterize langsages attested tua, they are probaly equally
true of languages spoken up to 10,000 years ago any Fate, TS
Sssumption il Se made i What flows
‘As an aksration of how ‘ypology can be used 10 conse
reconstructions, ne may consider 2 simple example Involving 2
Confit between the comparative method a sty applied, and
typological posit or paustlty. First, et us take an example
Where there no confi In both European and Syrian varices
‘fF Romany, «language (or more sccuraely group of languages)
Sf the Info-aryan subbrench of the Tndoteaian branch
Indo-Eeropean, there are two sibilant phonemes,» and fet if
Me look at correspondence ses betwcen European and Sian
Romany, we tind hat there ate tres, wilt oo evidence of
‘competency distribution betwees| any pair of them. Given
ths net applcstion of the comparative method requires seine6 “TvPOLOGY AND RECONSTRUCTION:
up three sibilant phonemes in Prote-Romany, ab in Table 3.1
(aatla. 1085: 245-6). Now, there is-nothing. typological
Iinplusible about hee stant phonemes ins singe language,
Indeed the sytem hypothesized for Proto Romany s exely the
‘ime as that attested in Sanit. Actually the compazion goes
further inthe ene: Sanit san ancent language ofthe Todo
‘Aryan sub-branch, and in cognates we find exact corespondence
isemeen. the. reconstructed Proto-Romany” slants andthe
tested siblans of Sanskt, so that here the Sonsuit data serve
ts independent confimation of te correciess ofthe reconstruc
tion of Proto Romany
sue 34 Romany sans
sopeon Romany ‘Siri Raman Prete Romany
: * %
i : %
A citferent situation is provided by vowel correspondences in
iinkchee and Korvak, two closely elated languages of the
(Chukotke-Kemchatkan family ofnor-castem Siberia. Chokchee
‘Sa decay relatively homogenous language. while Korsak is
Gialecally much more divided, tndeed some varieties that are
Sometimes considered. dialects of Korsak are perhaps. more
Sccurtely to be considered separate Tanguagess for example
‘Alutor~ Both Tanguoges ave a system of vowel hatnony
‘whereby the vowels are died ito two sets of thee such that
Al'the vowels 9 given wort most Belong to one sn only one
Sets in Words with fecessve vowels, these-are (1, e: in WOWdS
With dominant vowels, dhe corespondents ae € 0.0; e and
fre clearly distinguishable morpophonemial. In. adion,
‘ther es schva vowel a, which clearly sometimes Phoneme
‘Chulchee and posibly sometimes phonemic in Kory, although
the schwa wil play n role in the following discussion. [general
‘i given vowel of Chukchee appears also i cognate Koryak items,
{ee ha Some stances of ape a the prea mer
‘of such correspondences varying from dale wo dialect until one
Fesches Aluor, the varety most ditiact from Chukehec, in
‘which ale correspond toa indeed where the division between
the two set of vowels lowe completely so that there ate just
three (nowsehwa) vowels fm, a. Let vs suppose that we take
‘Ghukchee and three difereht varieties af Koryak with dlereat
(GENERAL CONSIDERATIONS 5
issibtions of and ad tht we have coeespondenes asin
{fale 32° Now? ste applieaon of the comparative method
oul que tt setup dor Prote-Chkotko Ramehatkan te
{otne! vowel phones one foreach sameapondene stim the
hot mist owe ew inl ore
Fypobpilly.sich'a sem te highly Inpausbi, copecally
en iat the reenrton fas no eonespondingvchness of
nck vone And indeed, ae fara lam aware no investigator of
{Chute and Rory his been led fo post sch astm for
Proto Chath Kemchatkan Rather, have sumed that We
have here as of eel ifesin,wih ue shige 10
‘oe domaip bucomee move extensive a one moves foie
(Deny tou Chee
Cicer Ronokh Konak? Konak
In the core of this article, we will examine two ses of dat
from the reconsriction of Proto-lado-European phonology. The
Choice of these datasets is aot acidenal. The protodanguoge of
the Indo-European family isthe one Uiat has been subjest 0
tnost intensive investigation within historeab-comparatve Lingus
Shame of di ae sl or tein tetera
‘methodologeal principles: yet despite (or pesbaps Becmse
the amount of work that hae been dove an Fecostructing Proto
Indo-European, there oman areas of reat, cOMleoversy, i
particular surounding the phonation types eg. voicing, spici-
on) of Protondo-Europenn and concerning. the relation
between the phonemically distinct vowel qualities andthe
laryngeal consonant phoremes of Prote Indo-European. The
‘choice of phonology rather than syntax, even though much ofthe
recent work in typology has been syntactic, follows ot so much
from any question of principle ae from the greater ease. with
which spose questions can he formulated and even, at east 10
Some evtent; answered. with respect to the tocomstraction of
Proto-lndo European phonology.